Alloy Outage History

The issue has been resolved.

Between March 4 at 11:00 PM ET and March 5 at 10:07 AM ET, we experienced an elevated rate of memory-related errors impacting evaluations running through JQ input attributes. As a result, some evaluations returned a null value for the input attribute instead of the correct value, potentially interrupting decisioning in the Workflow. Impacted applications would need to be re-run. Only workflows using input attributes with JQ were impacted; other Workflow logic was not impacted. The issue has been identified and resolved, and all systems are now operating normally.

We recently identified and resolved an issue affecting some plugins initialized through the Alloy SDK. For a subset of SDK Plugins, applications initializing the SDK may have experienced unexpected behavior, which could present as UI instability or premature close events. We mitigated the issue by rolling back a recent SDK change. Following the rollback, normal plugin behavior was restored. If you continue to experience any issues, please contact Alloy Support.

API functionality has been fully restored. We are continuing to investigate the impact and look into root causes.

Realtime webhook processing has been fully restored, and has maintained stability after monitoring. All new webhook events are now completing in realtime as expected. To further strengthen system stability, the engineering team has deployed an additional mitigation to prevent webhook queue backlogging under similar spike conditions. A thorough postmortem will be completed a Root Cause Analysis (RCA) can be shared with customers - please email [email protected] if you would like to receive the RCA. We are also investing in further improvements to the reliability and resiliency of our webhook queueing mechanism in the future to prevent event loss in the future. Please note, action is required to ensure applications process: During the incident, webhook events generated between 2:24 PM - 4:38 PM ET and 5:42 - 5:45 ET were not successfully processed and were permanently lost. Webhooks generated outside of this time window were not impacted. This issue affects applications that are stuck at a Journey step dependent on an Alloy webhook response to proceed (for example, action nodes), as well as applications that completed with a Manual Review outcome but are not reflecting back in core after review. Applications submitted during the affected window that are currently blocked due to a missing webhook dependency will need to be replayed. We recommend resubmitting those applications via API. Re-running an application will leverage cached data by default and will not trigger new calls to external data vendors.

The root cause has been identified. Response times for the Dashboard and API have returned to normal levels since 3:46 PM ET.